[NO-TICKET] Upgrade to libdatadog 14.1

[ivoanjo]

What does this PR do?

This PR upgrades the libdatadog dependency to the latest 14.1 release.

Motivation:

This PR fixes the issues we uncovered during testing of 14.0 in #4065: specifically, that triggering a stack overflow with native code would result in incorrect behavior (a segfault, not a stack overflow exception) when crashtracking was enabled.

Change log entry

Upgraded libdatadog dependency to version 14.1

Additional Notes:

N/A

How to test the change?

Our existing test coverage includes libdatadog testing.

The native code stack overflow is a bit of a pain to trigger because GCC seems to like optimizing away such things.

What I did was compile the profiler native extension with export DDTRACE_DEBUG=true (remember to run clean first!), and then adding the following trivial stack overflow:

diff --git a/ext/datadog_profiling_native_extension/profiling.c b/ext/datadog_profiling_native_extension/profiling.c
index 96637cd110..4ec5505147 100644
--- a/ext/datadog_profiling_native_extension/profiling.c
+++ b/ext/datadog_profiling_native_extension/profiling.c
@@ -38,6 +38,12 @@ static VALUE _native_enforce_success(DDTRACE_UNUSED VALUE _self, VALUE syserr_er
 static void *trigger_enforce_success(void *trigger_args);
 static VALUE _native_malloc_stats(DDTRACE_UNUSED VALUE _self);

+static VALUE blow_stack(DDTRACE_UNUSED VALUE _self) {
+  fprintf(stderr, ".");
+  blow_stack(Qnil);
+  return Qnil;
+}
+
 void DDTRACE_EXPORT Init_datadog_profiling_native_extension(void) {
   VALUE datadog_module = rb_define_module("Datadog");
   VALUE profiling_module = rb_define_module_under(datadog_module, "Profiling");
@@ -72,6 +78,8 @@ void DDTRACE_EXPORT Init_datadog_profiling_native_extension(void) {
   rb_define_singleton_method(testing_module, "_native_trigger_holding_the_gvl_signal_handler_on", _native_trigger_holding_the_gvl_signal_handler_on, 1);
   rb_define_singleton_method(testing_module, "_native_enforce_success", _native_enforce_success, 2);
   rb_define_singleton_method(testing_module, "_native_malloc_stats", _native_malloc_stats, 0);
+
+  rb_define_singleton_method(testing_module, "blow_stack", blow_stack, 0);
 }

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 97.73%. Comparing base (33f30f7) to head (0fa0e1f).
Report is 2 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #4082      +/-   ##
==========================================
+ Coverage   97.72%   97.73%   +0.01%     
==========================================
  Files        1338     1338              
  Lines       80251    80251              
  Branches     4017     4017              
==========================================
+ Hits        78425    78434       +9     
+ Misses       1826     1817       -9     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[lloeki]

GCC seems to like optimizing away such things

Sounds like TCO?

[ivoanjo]

Possibly, but I didn't check so I didn't want to make claims regarding that ;)

[pr-commenter]

Benchmarks

Benchmark execution time: 2024-11-07 10:39:36

Comparing candidate commit 0fa0e1f in PR branch ivoanjo/libdatadog14_1-upgrade with baseline commit 33f30f7 in branch master.

Found 0 performance improvements and 0 performance regressions! Performance is the same for 29 metrics, 2 unstable metrics.

[lloeki]

LGTM, thanks for all the testing on this!